Statistical significance based on the length and position of the local score in a model of i.i.d. sequences.

Agnes Lagnoux (1) , Sabine Mercier (1) , Pierre Vallois (2, 3)

1 IMT - Institut de Mathématiques de Toulouse UMR5219
2 BIGS - Biology, genetics and statistics
3 IECL - Institut Élie Cartan de Lorraine

The local score of a DNA sequence, also called Smith and Waterman score \citealp{Wat95}, is a statistical tool that allows to detect biologically interesting sequence regions. Firstly, we study the accuracy of classical results on the local score distribution using a Kolmogorov and Smirnov goodness of fit test. Secondly, we highlight how the length of the segment that realizes the local score influences the classical setting based on local score only. Finally, we measure with numerical simulation the difference between the classical local score and the new one we have introduced.

Domaines: Mathématiques [math]; Statistiques [stat]

Agnes Lagnoux, Sabine Mercier, Pierre Vallois. Statistical significance based on the length and position of the local score in a model of i.i.d. sequences.. Bioinformatics, 2017, 33 (5), pp.654-660. ⟨10.1093/bioinformatics/btw699⟩. ⟨hal-01301246v2⟩